www.sciencedirect.com
c o r t e x x x x ( 2 0 1 2 ) 1e1 8
Available online at
Journal homepage: www.elsevier.com/locate/cortex
Research report
Splitting of the P3 component during dual-task processing ina patient with posterior callosal section
Guido Hesselmann a,*, Lionel Naccache b,c,d,e, Laurent Cohen c,d,e and Stanislas Dehaene f,g,h
aVisual Perception Laboratory, Department of Psychiatry, Charite Campus Mitte, Berlin, GermanybAP-HP, Groupe Hospitalier Pitie-Salpetriere, Department of Neurophysiology, Paris, FrancecAP-HP, Groupe Hospitalier Pitie-Salpetriere, Department of Neurology, Paris, Franced INSERM, ICM Research Center, UMRS 975, Paris, FranceeUniversite Paris 6, Faculte de Medecine Pitie-Salpetriere, Paris, Francef INSERM, Cognitive Neuroimaging Unit, Gif sur Yvette, FrancegCEA, I2BM, NeuroSpin Center, Gif sur Yvette, FrancehCollege de France, Paris, France
a r t i c l e i n f o
Article history:
Received 19 October 2011
Reviewed 20 January 2012
Revised 8 February 2012
Accepted 18 March 2012
Action editor Ray Johnson
Published online xxx
Keywords:
Psychological refractory period (PRP)
Dual-task interference
Split-brain
Event-related potentials (ERPs)
P3
* Corresponding author. Department of PsycE-mail address: [email protected]
Please cite this article in press as: Hesselwith posterior callosal section, Cortex (2
0010-9452/$ e see front matter 2012 Elsevdoi:10.1016/j.cortex.2012.03.014
a b s t r a c t
When two concurrent sensorimotor tasks have to be performed at a short time interval, the
second response is generally delayed at a central decision stage. However, in patients who
have undergone full or partial transection of forebrain fibers connecting the two hemi-
spheres (split-brain), independent structures subserving all processing stages should reside
in each disconnected hemisphere, thus predicting parallel processing of dual tasks.
Surprisingly, this prediction is usually not verified behaviorally. We reasoned that brain
imaging with high-density recordings of event-related potentials (ERPs) could clarify the
extent and limits of parallel processing in callosal patients. We studied a patient (AC) with
posterior callosal section in a lateralized number-comparison task. Behaviorally, the split-
brain patient showed robust dual-task interference, superficially similar to the psycho-
logical refractory period (PRP) effect in the control group of 14 healthy subjects, but
significantly different in important aspects such as slowing of response times in the first
task. Analysis of ERPs revealed that the parietal P3 component became split into distinct
contralateral components in the patient, and was dramatically reduced for targets in his
left visual field. In contrast to the control group, P3 latencies showed minimal to nonex-
istent postponement related to dual-task processing in the patient. In summary, our
findings suggest that the left and right hemisphere networks normally involved in a single
distributed global neuronal workspace that underlies the generation of the P3 component
and serial processing, became strongly decoupled after a posterior callosal lesion.
2012 Elsevier Srl. All rights reserved.
hiatry, Charite Campus Mitte, 10117 Berlin, Germany.(G. Hesselmann).
mann G, et al., Splitting of the P3 component during dual-task processing in a patient012), doi:10.1016/j.cortex.2012.03.014
ier Srl. All rights reserved.
mailto:[email protected]/science/journal/00109452www.elsevier.com/locate/cortexhttp://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 82
1. Introduction
Fig. 1 e PRP paradigm. In dual-task trials, two target
numbers (here, 4 and 9) were presented on the screen for
100 msec, and subjects were instructed to perform two
successive number-comparison tasks (smaller or larger
than 5?). The SOA between the first (T1) and the second
target (T2) was varied between 100, 300, and 800 msec.
Subjects responded to both targets with manual button
presses. In two blocks of the dual-task condition, T1 was
either presented on the left and T2 on the right (T1LT2R
trials), or T1 was presented on the right and T2 on the
left (T1RT2L). In the single-task condition, only T1 was
presented, in two blocks with T1 either left (T1L) or right
of fixation (T1R).
1.1. Dual-task performance in split-brain patients
Even simple multi-tasking situations can reveal striking
capacity limits of human information processing (Marois and
Ivanoff, 2005). When neurologically normal subjects are asked
to perform two tasks simultaneously or in close succession,
severe interference is typically observed, although perfor-
mance may improve when the task-relevant items are
submitted to the relatively independent processing resources
of the cerebral hemispheres (Scalf et al., 2007; Banich, 1998).
Here, we address the issue of how patients with disconnected
hemispheres perform in such situations. A wealth of split-
brain studies suggests that separate and independent struc-
tures subserving all processing stages necessary for the
completion of simple sensorimotor tasks should reside in
each disconnected hemisphere (Gazzaniga, 1995, 2005).
Although dual-task interactions had been studied in split-
brain patients before [e.g., (Holtzman and Gazzaniga, 1982),
patient VP], Pashler and colleagues were the first to employ
a variant of the psychological refractory period (PRP) paradigm
in a group of four such patients (JW, NG, VP, LB) who had
undergone surgical transection of the corpus callosum for the
control of intractable epilepsy (Pashler et al., 1994). In the
classic PRP paradigm, two target stimuli (T1 and T2) are pre-
sented in brief succession, and subjects responses (R1, R2) to
both targets are recorded (Welford, 1952; Telford, 1931).
Response times to the second stimulus (RT2) have been shown
to exhibit a significant increase when the stimulus-onset
asynchrony (SOA) between the two tasks is shortened, while
response times to the first stimulus (RT1) remain largely
unaffected by SOA. In the study by Pashler and colleagues,
subjects were presented with a sequence of two stimuli in the
left and right visual hemifield, and responded with the left
hand to left-field stimuli and with the right hand to right-field
stimuli (two-alternative forced-choice of vertical position).
Thus, the input and output for each visual task were confined
to either isolated hemisphere. The authors hypothesized that
PRP effects should be eliminated in split-brain patients, under
the assumption that the interference depends upon direct
cortico-cortical connections between the two hemispheres.
Surprisingly, however, patients with commissurotomy
showed a strong dual-task slowing (PRP effect), which e
superficially at least e seemed highly similar to that of the
normal control group.
1.2. The PRP
Modern theories of the PRP, such as the central bottleneck
model, commonly involve three stages of processing:
a perceptual (P), a central (C), and a motor (M) stage (Pashler,
1994, 1998). According to the model, P and M stages can
occur in parallel and may overlap for two tasks, while the C
stage is strictly serial. Thus, at short SOAs, central processing
for T2 is deferred until central processing for T1 is completed,
and RT2 is increased [e.g., (Sigman and Dehaene, 2005), Fig. 1].
It has been proposed that response selection, i.e., themapping
between sensory information and motor action, forms
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
a structural bottleneck and underlies serial processing at the C
stage (Pashler and Johnston, 1989; De Jong, 1993). Pashler and
colleagues discussed their findings from split-brain patients
within the framework of the bottleneck model, concluding
that intact subcortical structures might participate in sched-
uling multiple stimulus-response sequences in callosum-
sectioned patients. The exact nature of this coordination,
however, remained to be elucidated (Pashler et al., 1994).
In a related study on motor control, it was shown that
a split-brain patient (JW) could plan and produce incompatible
bimanual movements without dual-task interference, thus
arguing against the claim of an intact unitary response
selection bottleneck in callosotomy patients (Franz et al.,
1996). In a series of single-case studies on the same patient,
Ivry and colleagues examined this apparent contradiction
using variations of the PRP task (Ivry et al., 1998; Ivry and
Hazeltine, 2000). Their results showed robust dual-task slow-
ing in the split-brain patient, as well as in a group of control
participants. Yet, the data also revealed significant differences
between the patients and the normal groups dual-task
performance under different stimulus-response mappings,
indicating that the dual-task slowing in callosal patients can
be accounted for by a bottleneck associated with response
initiation rather than response selection (Ivry et al., 1998).
A recent study argues that the apparent PRP effects in split-
brain individuals are the consequences of a prioritization
strategy adopted by the subjects when performing the two
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 3
tasks (Hazeltine et al., 2008). In a standard PRP experiment,
although the experimenter intends the subject to respond as
fast as possible to each target, the task structure and
instructions might be interpreted by subjects as a need to
always respond first to the first target, and second to the
second one. Performing such a serial task might then be
disproportionately difficult in callosal patients whose two
hemispheres cannot readily exchange information about
which stimulus came first. It would thus result in a conserva-
tive strategy which could severely slow down both responses.
Indeed, it has been shown in normal subjects that dual-task
interference can be strongly modulated by instructions
about task priorities (Schumacher et al., 2001). In their study,
Hazeltine and colleagues tested two split-brain patients (JW,
VP) in a dual-task paradigm in which lateralized stimuli for
the two tasks were presented simultaneously, and patients
were instructed to simply respond to the stimuli as quickly as
possible. In disagreement with previous studies (Pashler et al.,
1994; Ivry and Hazeltine, 2000; Ivry et al., 1998), patients
showed much smaller or nonexistent dual-task costs, which
suggests that they were able to simultaneously select
responses for the two hands without central processing
limitations.
1.3. Event-related potential (ERP) studies of the PRP
All of the above dual-task studies relied on behavioral
measures. To further explore the issue of dual-task processing
with partially disconnected hemispheres, we sought to
describe the neural substrates of dual-task costs and inter-
hemispheric interference in a partial split-brain patient. To
that aim, we recorded high-density electroencephalography
(EEG) data in a patient (AC) with posterior callosal section in
a lateralized PRP task. Behavioral and EEG data from our
previous work on the same task (Hesselmann et al., 2011)
served as healthy control group data. Our study of ERPs
focused on the sensory N1 and the post-sensory P3 compo-
nents, using a linear regression method optimally suited for
the analysis of single-trial data.
A number of previous ERP studies investigating the PRP
effect have targeted the amplitude and latency of the P3(b)
component, which is characterized by a positive deflection
broadly distributed over the scalp, with a focus over parietal
electrodes (Picton, 1992). The P3 has been associated with post-
perceptual processes such as the context-updating of working
memory (Donchin and Coles, 1988; Donchin, 1981), decision-
related processing (Verleger et al., 2005), and the access of
a target stimulus to a global neuronal workspace necessary for
conscious report (Sergent et al., 2005; Del Cul et al., 2007).
Previous dual-task investigations have provided evidence for
a sensitivity of P3 amplitude to dual-task interference (Isreal
et al., 1980; Kok, 2001). Based on the observation that P3
latencies showed significant postponement directly propor-
tional to the PRP effect, some studies have proposed that the
P3(b) component primarily indexes the central cognitive
processes mediating the PRP effect (DellAcqua et al., 2005;
Sigman and Dehaene, 2008; Hesselmann et al., 2011).
Evidence from other studies, however, occasionally shows
large discrepancy between RT2 and T2-P3 latencymodulations,
suggesting at least partially independent sources for PRP and P3
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
effects (Luck, 1998; Arnell et al., 2004). The latencies of earlier
sensory ERP components, such as the P1 and N1, have been
consistently reported to remain stimulus-locked to both targets
and show no postponement related to the PRP (Brisson and
Jolicur, 2007; Sigman and Dehaene, 2008).
In this context, the main question of our study was
whether the P3 responses would show a PRP delay in a callosal
patient, in a paradigm that systematically reveals such a delay
in normal subjects (Hesselmann et al., 2011). Based on the
previously mentioned behavioral study by Hazeltine et al.
(2008), we predicted that the P3 wave might indicate a great
degree of parallel processing in our partial split-brain patient,
even if the behavioral responses superficially suggest a dual-
task delay.
2. Materials and methods
2.1. Participants
The patient (AC) tested in this study was a 36-year-old male
right-handed native French speaker. He obtained the
baccalaureat (French high school graduation) and completed
two years of university studies. Following surgery for a cere-
bral hemorrhage in his left mesial parietal lobe due to a small
arteriovenous malformation, he presented a partial section of
the posterior half of the corpus callosum [for more details and
anatomical images, see (Intriligator et al., 2000; Cohen et al.,
2000)]. A magnetic resonance imaging (MRI) examination
revealed a porencephalic cyst in the left parietal lobe (Frak
et al., 2006) and a cut of the posterior half of the corpus cal-
losum (Michel et al., 1996), from the posterior midbody (III) to
the splenium (V), according to the tractography-based par-
cellation by Hofer and Frahm (2006). At the time of the study in
April 2008, approximately 14 years after surgery, ACworked as
an accountant, had normal visual fields, no signs of optic
ataxia, no paresis or hypesthesia, no simultanagnosia or
visual neglect, and no apraxia. The experimentwas conducted
at the Department of Neurology at the Pitie-Salpetriere
hospital in Paris, France. The healthy sample consisted of 14
male right-handed native French speakers who participated
in a previously published EEG-functional MRI (fMRI) experi-
ment (Hesselmann et al., 2011), which was conducted at the
NeuroSpin neuroimaging center in the CEA campus of Saclay,
France. Behavioral data from this study were re-analyzed in
the current study. To increase statistical power, we included
behavioral data from two subjects which had been excluded
previously based on neuroimaging data. All healthy subjects
(mean age 23, range 19e28 years) had normal or corrected-to-
normal vision. All participants provided informed written
consent to take part in the experiment.
2.2. Design and procedure
The experimental design was identical to that from our
previous study (Hesselmann et al., 2011). The patient and the
control subjects were asked to perform two number-
comparison tasks (smaller or larger than 5?) on two succes-
sive digits presented left or right of fixation. Smaller-larger
magnitude comparison has previously been shown to be
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 84
spared for left visual field (LVF) and right visual field (RVF)
targets in split-brain patients (Colvin et al., 2005; Seymour et al.,
1994) as well as in a patient with posterior callosal lesion
(Cohen andDehaene, 1996), when quantities are represented in
form of Arabic digits. All subjects were instructed that they had
to respond accurately and as fast as possible to each of them,
according to the order of their appearance. The target stimuli
(numbers 1, 4, 6, or 9) were presented in white font (Courier
New) on a black background for 100 msec (Fig. 1) using Eprime
software (Version 1.1, Psychology Software Tools Inc., USA).
The SOA between the first (T1) and the second target (T2) was
varied between 100, 300, and 800 msec. The size and eccen-
tricity of the target stimuli were carefully chosen so that
subjects could perform the number-comparison task while
maintaining fixation; we observed only small (
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 5
brief, multi-channel time-series of EEG data (spatio-temporal
templates) for each ERP component of interest (N1, P3) were
extracted from the single-task condition in which these
components did not overlap. Multiple linear regression was
then used to project, for each time point, the EEG to the
previously defined component templates. The resulting time-
series of parameter estimates (beta values) quantify, for each
time point, to what extent ERP component activity compa-
rable to the single-task condition was present in the dual-task
condition. Note that the regressionmethod provides powerful
denoising of the EEG signal thus making it well suited for the
statistical analysis of single-trial data in our single-case study.
The following paragraphs provide a more detailed description
of the method.
First, we defined spatio-temporal templates which were
designed to capture the temporal and scalp distribution
characteristics of the N1 and P3. Two templates were used for
the components evoked by left and right single targets (T1L
and T1R), respectively. Estimation of the onsets of the N1 and
P3 components was based on the time course of the global
field power (GFP) (Lehmann and Skrandies, 1980). As can be
seen in Fig. 4A, a first GFP maximum occurred at approxi-
mately 168 msec, and a second GFP maximum started at
Fig. 2 e Behavioral PRP results. Average RTs at SOAs 100, 300, an
solid lines) and to the second target (RT2, dashed lines), separ
squares, respectively). (A) Split-brain patient data (N [ 1). (B)
represent standard error of the mean (SEM). Upper and lo
respectively.
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
approximately 328msec. The corresponding ERP topographies
at 168 msec revealed the posterior negativity of the lateralised
T1L-N1 and T1R-N1 components (Fig. 4B). The ERP topography
at 328 msec revealed the parietal positivity of the T1L-P3 and
T1R-P3. In contrast to the data from the healthy subjects, the
P3 components appeared to be as lateralised as the N1
components. The smaller peak at 112 msec corresponded to
the sensory P1 component which was not analyzed further.
Intermediate and later GFP peaks could not be associated with
known ERP components. The spatio-temporal distributions
were extracted from the EEG data based on the corresponding
peaks, as follows: for the N1, we used the voltages in
a 100 msec window centered on 168 msec after the single
target onset, thus from 118 to 218 msec; similarly, for the P3,
we used the voltages in a 200 msec window centered on
328 msec (i.e., from 228 to 428 msec after single target onset),
thus non-overlapping with the N1 time window. Note that we
chose a longer template for the P3 to account for the broader
maximum of this component.
Next, for each time point of the EEG, we used a multiple
linear regression procedure on sliding windows of data to
extract the single-trial temporal profiles of the four spatio-
temporal components (LVF-N1, RVF-N1, LVF-P3, RVF-P3).
d 800 msec are shown for responses to the first target (RT1,
ately for congruent and incongruent trials (filled and open
Data from the healthy control group (N [ 14). Error bars
wer panels show data from T1LT2R and T1RT2L trials,
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 86
High beta weights, for a given time point, indicate a good fit
between the EEG data and the spatio-temporal template
within the corresponding time window. Single-trial beta
weights are presented as ERP-images, as provided by the
EEGLAB software (Figs. 4D, 7 and 8). These plots are color-
coded (red: positive values, green: zero, blue: negative). For
better visualization, single-trial beta weights were smoothed
across trials with a 20-trials widemoving rectangular window.
Finally, the extracted single-trial beta weights were
submitted to statistical analysis. To test for significant devia-
tions from zero, beta weights in all dual-task conditions were
submitted to one-sample, one-tailed permutation t-tests with
correction for multiple comparisons (Blair and Karniski, 1993),
using the Mass Univariate ERP Toolbox developed at the
Department of Cognitive Science at the University of California,
San Diego (Groppe et al., 2011).We used 5000 permutations to
estimate the distribution of the null hypothesis, and an alpha-
level of .005 to obtain the critical t-scores (tmax). In Figs. 3, 4C, 7
and 8, gray horizontal bars above the x-axis indicate significant
time points; points within a consecutive series of at least 10
significant time points (40 msec) are color-coded according to
condition. To test for differences between beta weights in
different conditions, we extracted single-trial beta weights for
time windows centered on the average peaks of the corre-
sponding components. N1 betas were extracted over a time
window of 24 msec around the peak time; P3 betas wereextracted over a time window of 48 msec around the peaktime. In this way, we obtained, for each trial and each condi-
tion, a beta value for N1 and P3. The single-trial beta values
were submitted to repeated-measures ANOVA with factors
Fig. 3 e Average patient ERPs relative to T1 onset at parietal elec
T1LT2R trials (middle panels), and dual-task T1RL2L trials (rig
points; significant time points within a consecutive series of a
according to condition. Small vertical bars on the x-axis indic
trials. For display purposes, the ERP data have been subjected t
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
SOA and hemisphere. Analysis of beta weights using non-
parametric randomization tests (Todman and Dugard, 2001)
instead of repeated-measures ANOVA yielded the same
statistical results as reported in our paper.
3. Results
3.1. Behavioral results: PRP effect
Table 1 provides a summary of patient ACs response accu-
racies and reaction times (RTs). In single-task trials, patient
ACs response accuracy rates were comparable to those
observed in the healthy sample [94.3% vs 96.3 2.67%;mean SD; t13 < 1]. In dual-task trials, they were slightly butsignificantly lower than in the healthy group (86.0% vs
94.4 3.21%; t13 2.53, p .013), and comparable for the first(87.7%) and second responses (84.4%). The patients response
accuracies were only minimally modulated by SOA and by the
laterality of target stimuli. Error trials were removed from all
further analyses (behavioral, EEG).
In the single-task condition, the patients mean RT was
significantly shorter for LVF targets (T1L: 516 67 msec;mean SD) than for RVF targets (T1R: 653 157 msec;F1,86 59.23, p < .001, h2 .41). As can be seen in Fig. 2A,a similar effect of visual field was observed in dual-task trials:
RT1 was 677 146 msec for LVF targets in T1LT2R trials and756 239 msec for RVF targets in T1RT2L trials (F1,217 18.73,p < .001, h2 .079); RT2 was 833 216 msec for LVF targets inT1RT2L trials and 906 215 msec for RVF targets in T1LT2R
trodes P3 and P4, in single-task trials (left panels), dual-task
ht panels). Gray horizontal bars indicate significant time
t least 10 significant time points (40 msec) are color-coded
ate T2 onset at SOA 100, 300, and 800 msec in dual-task
o a 20 Hz low-pass filter.
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Fig. 4 e Analysis of ERPs in single-task trials. (A) Average GFP of ERPs in T1R and T1L trials. GFP peaks were used to identify
ERP components P1, N1, and P3 (gray rectangles). Intermediate and later peaks could not be associated with known ERP
components. The black rectangle on the x-axis represents the on- and offset of the stimulus. (B) The 2D scalp
topographies show the ERP topographies evoked by right targets (T1R) and left targets (T1L) at six different time points
(112 [ P1, 168 [ N1, 236, 328 [ P3, 428, and 528 msec). Small circles represent single electrodes; Pz is the fifth midline
channel from the bottom. (C) Results of the multiple regression method applied to single-task trials. Beta weights for
T1-N1 (green, red) and T1-P3 (blue, black) are plotted against time from T1 onset. Horizontal bars above the x-axis
indicate significant time points (deviation from zero). (D) Plot of single-trial beta weights in T1R and T1L trials, color-
coded (red: positive, green: zero, blue: negative). Ticks on the y-axis are every 50 trials.
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 7
trials (F1,217 4.20, p .042, h2 .02). In healthy subjects, visualfield did not modulate RT (Fig. 2B). Overall RTs in single-task
trials (534 20 msec; mean SE), and RT1 in dual-task trials(651 22msec) were comparable to the patients performance(t13 < 1), while RT2 was significantly shorter in the control
group (666 24 msec; t13 2.19, p .024).Fig. 2A illustrates that, in both trial types, the patients RT2
increased significantly with decreasing SOA, in accordance
with the classical PRP model (Pashler, 1994; Pashler and
Johnston, 1989). RT1 showed a smaller, yet consistent and
significant increasewith decreasing SOA. In T1LT2R trials, RT2
increased by 491msec as SOA decreased from 800 to 100msec,
while RT1 increased by 203 msec over the same time range.
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
In T1RT2L trials, the corresponding values were 531 msec and
205 msec. The F ratios and partial eta squared values under-
line the difference in magnitude between the main effects of
SOA for RT1 and RT2, both in T1LT2R trials (RT1: F2,66 29.95,p < .001, h2 .48, .66; RT2: F2,66 91.89, p < .001, h2 .74, .84) and in T1RT2L trials (RT1: F2,72 13.68, p< .001, h2 .28, .96; RT2: F2,72 107.77, p < .001, h2 .75, .74).
Taking only RTs at SOAs 100 and 300 into account (i.e., SOAs
within the interference regime), the RT2 slopes were 1.56 forT1LT2R trials and 1.41 for T1RT2L trials, thus larger than thetheoretical slope of1 predicted by the classical PRPmodel forshort SOAs. In the healthy sample, the corresponding RT2
slopeswere.86 .06 (mean SE) and.96 .07, respectively
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Fig. 5 e Correlation between T1R and T1L single-task ERP topographies. The solid lines represent the moment-to-moment
Pearson correlation coefficients (r) between ERP topographies evoked by T1R and T1L targets in single-task trials (red:
patient data, N [ 1; black: healthy control group data, N [ 12). Separately for the controls (upper panels) and the patient
(lower panels), topographies in T1R and T1L trials are shown for time points ta (controls: 200 msec, patient: 168 msec), tb(controls: 380 msec, patient: 328 msec), and tc (controls: 580 msec, patient: 580 msec). The topographies are 2D
representations of the different 3D electrode layouts (patient: 256 electrodes, Pz is the fifth midline channel from the bottom;
controls: 62 electrodes, Pz is the third midline channel from the bottom). To reduce noise in the patient data, the correlation
values were smoothed with a moving average filter (length: 15 data points, 60 msec). The gray shading represents SEM.
c o r t e x x x x ( 2 0 1 2 ) 1e1 88
(Fig. 2B). The test for a difference in RT2 slopes between the
patient and the healthy sample yielded significant results for
both trial types (T1LT2R: t13 3.04, p .005; T1RT2L: t13 1.77,p .049). In T1LT2R trials, RT1 slopes at short SOAs weresignificantly larger for the patient than in the healthy sample
(.88 vs .15 .05, t13 3.82, p .001), but not in T1RT2L trials(.61 vs .30 .06, t13 1.28, p .112).
In a next step, we analyzed RT1-RT2 correlations across
SOAs. If central T2 processing is indeed postponed by central
processing of T1, then RT1 and RT2 should be more strongly
correlated at shorter SOAs, since the slowing in the first task is
propagated onto the second task (Pashler, 1994). In T1LT2R
trials, the Pearson correlation coefficient for patient AC
dropped from .76 at SOA 100 and .31 at SOA 300 to .01 at SOA
800; in T1RT2L trials, the coefficients were .80, .81, and .26,
respectively (Supplementary Fig. 2). Overall, there appeared to
be a good correspondence between RT1-RT2 correlations in
patient AC and in the healthy sample; t-tests for all SOAs
yielded no consistent differences between correlation coeffi-
cients in T1LT2R trials (t13 .52, p .306; t13 2.23, p .022;t13 1.20, p .126), and in T1RT2L trials (t13 .39, p .352;
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
t13 1.15, p .135; t13 .55, p .296; not corrected for multiplecomparisons).
3.2. Behavioral results: crosstalk effects
In our dual-task paradigm, the comparison between
congruent and incongruent trials allowed for the analysis of
crosstalk between T1 and T2 processing. Crosstalk is defined
as an effect of the congruency of the two target responses
(here, both larger or both smaller than 5) on RT1 and RT2,
respectively. The dependence of RT1 on the response that is
required for the second stimulus is referred to as backward
crosstalk (as opposed to crosstalk from T1 on RT2). Backward
crosstalk effects, which are usually observed when both
tasks are similar [(Logan and Delheimer, 2001; Logan and
Schulkind, 2000), but see (Miller, 2006)], appear to be in
disagreement with the strictly serial bottleneck model, and
have been interpreted as supporting central resource sharing
models (Navon and Miller, 2002; Tombu and Jolicur, 2003)
whereby the two tasks are performed partially in parallel and
with continuous variable relative priorities. They remain
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Fig. 6 e Individual P3 topographies for all EEG control subjects. Shown are the LVF-P3 (i.e., P3 evoked by targets in the LVF)
and RVF-P3 at 380 msec relative to T1 onset. Note that, in control subjects, EEG data were recorded simultaneously with
fMRI, and that EEG data quality is usually lower in such combined EEG-fMRI setups due to the interference between the
two recording methods. The topographies are 2D representations of the 3D electrode layout (62 electrodes, Pz is the third
midline channel from the bottom).
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 9
compatible, however, with strictly serial processing of the two
decisions, only with a partial leakage of sensory evidence
from one target on the decision concerning the other. The
leakage account builds upon the notion that the serial pro-
cessing bottleneck can be characterized as the accumulation
of evidence toward a decision boundary (Sigman and
Dehaene, 2005). In the current number-comparison para-
digm, the possibility of leakage of evidence is made more
likely by the finding that unattended and even subliminal
digits can automatically access a representation of their
magnitude (Naccache and Dehaene, 2001; Dehaene et al.,
1998; Sackur et al., 2008).
Visual inspection of Fig. 2A reveals that crosstalk and back-
ward crosstalk appeared to be present at SOA 100 in T1RT2L
trials (lower panel), but were virtually absent in T1LT2R trials
(upperpanel). Basedonourpreviousfindings (Hesselmannet al.,
2011), we restricted the analysis of crosstalk effects to the
shortest SOA, and calculated repeated-measures ANOVAs with
factors congruency and T1 laterality, separately for RT1 and
RT2data.Themaineffect congruencywasnot significant (RT1:
F1,33 2.64, p .114, h2 .07; RT2: F1,33 1.43, p .241, h2 .04).The congruency laterality interaction turned out to be
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
significant for crosstalk in RT2 (F1,33 5.37, p .027, h2 .14) butnot for backward crosstalk in RT1 (F1,33 1.90, p .177,h2 .06).Explorative analysis of backward crosstalk in T1RT2Ltrials revealed only a marginally significant effect (F1,38 3.25,p .079, h2 .08).
Fig. 2B shows that, in the healthy sample, neither crosstalk
effect was modulated by the laterality of the first target (all
Fs < 1); in both trial types, crosstalk from T1 to T2 in the RT2
data was larger than backward crosstalk from T2 to T1 in the
RT1 data, and crosstalk was more pronounced at the shortest
SOA, as described in more detail in our previous study
(Hesselmann et al., 2011). Finally, we directly compared
crosstalk at SOA 100 between the patient and the healthy
control group. In T1LT2R trials, crosstalk from T1 to T2 was
larger in the control group (t13 2.37, p .017), but we foundno significant difference for backward crosstalk (t13 .36,p .363). In T1RT2L trials, differences between crosstalkeffects in the split-brain patient and control group were not
significant (t13 1.05, p .157; t13 1.43, p .088).In sum, the PRP effect was only superficially present in the
callosal patient (SOA effect on RT2, RT1-RT2 correlations), but
it lacked important characteristics, as attested by significant
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Fig. 7 e Analysis of sensory ERP components in dual-task trials. (A) T1-N1 responses. The upper panels show average beta
time courses in T1RT2L and T1LT2R trials, separately for SOA 100 (red), SOA 300 (green), and SOA 800 (blue). The upper left
panel shows T1-N1 activity in the left hemisphere, the upper right panel shows T1-N1 activity in the right hemisphere. In
c o r t e x x x x ( 2 0 1 2 ) 1e1 810
Please cite this article in press as: Hesselmann G, et al., Splitting of the P3 component during dual-task processing in a patientwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 11
differences between the patients and the control groups
behavioral data (SOA effect on RT1, RT slopes, crosstalk).
3.3. EEG results: ERPs in single-task and dual-task trials
Fig. 3 shows the patients average ERPs in single-task (T1L, T1R)
and dual-task (T1LT2R, T1RT2L) trials at parietal electrodes P3
and P4. As can be seen, the N1 remained stimulus-locked to the
first and second target, respectively, over both hemispheres.
The P3 component appeared to bemuch smaller at electrode P4
than at electrode P3 in single-task trials, and was virtually
absent at electrode P4 in dual-task trials. Over the left hemi-
sphere (at electrode P3), the latency of the P3 component
evoked by the second target showed only small variation with
SOA. These observations will be analyzed in more detail using
multiple linear regression in the following sections.
3.4. EEG results: single-task (N1 and P3)
Fig. 4B illustrates that the ERP topographies evoked by targets in
T1L and T1R trials revealed a strikingly long-lasting lateraliza-
tion of activity. In contrast to the EEG data from the healthy
subjects, where a single P3 component with broad parietal
topography was evoked by both LVF and RVF targets
(Hesselmann et al., 2011), even at the time of maximal P3
activity (328 msec) the topography observed in the patient
appeared to be as contralateral as at the peak time of the
sensory P1 (112 msec) and N1 (168 msec). To corroborate this
result, Fig. 5 plots, as a function of time from T1 onset, the
moment-to-moment correlation between the topographies
observed in T1L and T1R trials, separately for the patient and
the controls. Correlation coefficients indicate the degree of
similarity between the topographies. Already at the time of the
N1 component (ta: 175e225 msec), ERPs showed evidence of
inter-hemispheric transfer in normals, as attested by a positive
correlation of topographies and the presence of an ipsilateral
N1, as previously reported [e.g., (Johannes et al., 1995)]. By
contrast, the patients ERPs showed almost inverse topogra-
phies in T1L and T1R trials with a strictly contralateral N1,
resulting in negative correlation coefficients in the same time
range (Cohen et al., 2000). For the controls, the coefficients
indicated a second convergence toward a shared topography
during the time of the central P3 (tb: 300e400 msec), which
remained stable until a later time period (tc: 550e650 msec). In
the patient, however, topographies did not converge toward
the left hemisphere, N1 can be detected at all SOAs (latency: 168
all SOAs (latency: 172 msec). Horizontal bars above the x-axis
lower panels show plots of single-trial beta weights in T1RT2L
blue: negative). T1-N1 activity can reliably be detected on a sin
every 50 trials. (B) T2-N1 responses. The upper panels show
separately for SOA 100 (red), SOA 300 (green), and SOA 800 (b
hemisphere, the upper right panel shows T2-N1 activity in the
N1 can be detected at all SOAs (latencies: 272, 472, and 972 m
detected as well, but the N1 peak at SOA 300 does not reach s
bars above the x-axis indicate significant time points (deviatio
beta weights in T1LT2R and T1RT2L trials, color-coded (red:
reliably be detected on a single-trial level in both hemispheres.
100, 300, and 800 msec. Ticks on the y-axis are every 50 trials.
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
a shared pattern during the time of the P3 (tb), but only at
amuch later time ofw600msec after target onset (tc). Thus, the
patients P3was split into two distinct P3s, one evoked by LVF
targets, the other evoked by RVF targets. Further analysis of
individual subject P3 topographies in healthy controls revealed
that none of the control subjects showed a lateralization of the
P3 comparable to that of the patient (Fig. 6). As a consequence
of this finding, we used four spatio-temporal profiles (LVF-N1,
RVF-N1, LVF-P3, RVF-P3) when applying the multiple linear
regression procedure to the patient data. Note that we refer to
ERP components evoked by LVF targets as LVF-N1 and LVF-P3,
and to components evoked by RVF targets as RVF-N1 and
RVF-P3.
Fig. 4C shows the results of themultiple regressionmethod
applied to single-target trials. Note that the definition of the
component profiles was based on these data, and that we
performed this analysis primarily to verify the capacity of the
multiple regression method to separate neural events
unfolding over time. In both T1R and T1L trials, betas for the
contralateral N1 and P3 show significant peaks at the expected
latencies of 168 and 328 msec, respectively. In the healthy
control group, N1 and P3 beta peaks were observed at 200 and
496msec, respectively (Hesselmann et al., 2011), but P3 related
activity started already at approximately 400 msec. The time-
series of beta values related to the ipsilateral N1 and P3
components, on the other hand, show no significant peaks.
This finding confirms that all four ERP components could be
reliably separated in the EEG data of the single-task condition.
Fig. 4D illustrates the same results as color-coded single-trial
data plots, and confirms the successful separation of neural
events at the single-trial level. However, there appears to be
a strikingdifferencebetweenpost-sensoryprocessing in the left
and right hemisphere. While the RVF-P3 activity in T1R trials
(i.e., in the left hemisphere) is temporally well-confined and
reliably detectable in almost all trials, the LVF-P3 activity in T1L
trials (i.e., in the right hemisphere) appears to be much less
reliable and not present in all trials. Based on this result, we re-
analyzed the EEG data in order to further optimize all artifact
correction methods, but the LVF-P3 beta results remained
virtually the same.
3.5. EEG results: dual-task (N1)
In the next step, multiple regression analysis was used to
parse the event-related neural activity in dual-task trials.
msec). In the right hemisphere, N1 appears to be smaller at
indicate significant time points (deviation from zero). The
and T1LT2R trials, color-coded (red: positive, green: zero,
gle-trial level in both hemispheres. Ticks on the y-axis are
average beta time courses in T1LT2R and T1RT2L trials,
lue). The upper left panel shows T2-N1 activity in the left
right hemisphere. In the left hemisphere, stimulus-locked
sec). In the right hemisphere, stimulus-locked N1 can be
ignificance (latencies: 260, 468, and 972 msec). Horizontal
n from zero). The lower panels show plots of single-trial
positive, green: zero, blue: negative). T2-N1 activity can
Small vertical bars on the x-axis indicate T2 onset at SOA
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Fig. 8 e Analysis of central ERP components in dual-task trials. (A) T1-P3 responses. The upper panels show average beta
time courses in T1RT2L and T1LT2R trials, separately for SOA 100 (red), SOA 300 (green), and SOA 800 (blue). The upper left
panel shows T1-P3 activity in the left hemisphere, the upper right panel shows T1-P3 activity in the right hemisphere.
c o r t e x x x x ( 2 0 1 2 ) 1e1 812
Please cite this article in press as: Hesselmann G, et al., Splitting of the P3 component during dual-task processing in a patientwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Table 1 e Behavioral results (patient AC) in dual-tasktrials. Average response times for T1 (RT1) and T2 (RT2)as well as SDs in msec. RT1 in T1L single-task trials:516 67 msec (91.7% accuracy). RT1 in T1R single-tasktrials: 653 157 (96.9% accuracy).
SOA 100 SOA 300 SOA 800
T1LT2L trials
RT1 803 262 627 105 600 72RT1 congruent 795 250 621 110 597 63RT1 incongruent 811 275 634 100 603 81RT2 1175 275 860 197 684 172RT2 congruent 1192 260 852 188 672 183RT2 incongruent 1158 290 868 209 696 163Accuracy task 1 89.6% 87.5% 80.2%
Accuracy task 2 78.1% 83.3% 83.3%
r(RT1, RT2) .71 .36 .01
T1RT2L trials
RT1 865 270 744 286 660 161RT1 congruent 810 269 738 301 663 174RT1 incongruent 920 262 733 274 656 149RT2 1104 290 822 247 573 112RT2 congruent 1022 309 827 271 549 116RT2 incongruent 1185 243 816 222 596 104Accuracy task 1 92.7% 87.5% 88.5%
Accuracy task 2 90.6% 87.5% 83.3%
r(RT1, RT2) .80 .81 .26
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 13
Fig. 7A (upper panels) shows that, as in the healthy control
group, the N1 evoked by the first target (T1-N1) was reliably
recovered for all SOAs in T1LT2R and T1RT2L trials, peaking at
approximately 168 msec after T1 onset. Positive beta weights
for T1-N1 in the left hemisphere were significantly larger than
in the right hemisphere (F1,91 25.39, p < .001, h2 .22); themain effect SOA was not significant (F < 1), and we found no
significant hemisphere SOA interaction (F2,182 1.63,p .198, h2 .02, .99). T1-N1 betas for RVF targets werepreceded by negative betas peaking at approximately
100 msec, which very likely are related to P1 activity (see RVF-
N1 betas in T1R trials, Fig. 4C). Lower panels in Fig. 7A,
showing single-trial beta weights, confirm that T1-N1 activity
could reliably be detected at the expected latencies in both
hemispheres on a trial-by-trial basis. Inspection of raw
averaged ERP amplitudes also revealed smaller T1-N1
components in the right hemisphere, while T1-P1 amplitudes
were comparable in both hemispheres (Fig. 3). This result
suggests that the finding of smaller N1 betas in the right
In the left hemisphere, P3 can be detected at all SOAs (latency:
activity appears to be absent. Horizontal bars above the x-axis
lower panels show plots of single-trial beta weights in T1RT2L
blue: negative). T1-P3 activity can reliably be detected on a sin
axis are every 50 trials. (B) T2-P3 responses. The upper panel
trials, separately for SOA 100 (red), SOA 300 (green), and SOA 8
left hemisphere, the upper right panel shows T2-P3 activity in
detected at all SOAs (latencies: 440, 632, and 1144 msec). In th
to be absent. Horizontal bars above the x-axis indicate signifi
show plots of single-trial beta weights in T1RT2L and T1L
negative). T2-P3 activity can be only identified in the left hemi
at SOA 100, 300, and 800 msec. Ticks on the y-axis are every 50
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
hemisphere was not simply due to a mismatch between the
spatio-temporal templates and the dual-task EEG data.
Fig. 7B (upper panels) illustrates that the N1 evoked by the
second target (T2-N1) remained stimulus-locked in dual-task
trials as well, as previously reported for normal subjects
(Hesselmann et al., 2011; Sigman and Dehaene, 2008). Similar
to the T1-N1, the second N1 was significantly larger in the left
than in the right hemisphere (F1,91 21.82, p < .001, h2 .19),while the main effect SOA turned out to be not significant
(F < 1). The ANOVA yielded a significant hemisphere SOAinteraction (F2,182 5.12, p .007, h2 .05, .93), due to thefact that the T2-N1 at SOA 300was comparably large in the left
hemisphere, and small in the right hemisphere. Lower panels
in Fig. 7B, showing single-trial beta weights, provide further
evidence for the stimulus-locking of the T2-N1, as well as for
the amplitude difference between hemispheres. Finally,
a comparison betweenT1-N1 and T2-N1 betaweights revealed
no significant difference (F < 1).
3.6. EEG results: dual-task (P3)
Fig. 8A (upper panels) shows that, in dual-task trials, significant
T1-P3 activity could only be observed in the left hemisphere,
peaking at approximately 328 msec, across all SOAs. This
finding is confirmed by single-trial beta weights in T1LT2R and
T1RT2L trials (Fig. 8A, lowerpanels). Therefore,we restricted the
ANOVA to EEG data from the left hemisphere. T1-P3 activity in
the left hemisphere appeared to be largest for SOA 800, yet the
main effect SOA did not turn out to be significant (F2,184 1.75,p .176, h2 .02, .99). Inspection of raw ERP amplitudes inparietal electrodes revealed that, in the patients right hemi-
sphere, P3-like activity is present only in a number of trials at
SOA100 (datanot shown). Thus, theabsenceof significantT1-P3
beta weights for the right hemisphere cannot be accounted for
by amismatchbetween the spatio-temporal template anddual-
task EEG data. Analysis of ERP responses in more centrally
located electrode clusters yielded similar results.
Similar to the T1-P3 findings described above, significant
T2-P3 activity could only be observed in the left hemisphere
(Fig. 8B, upper panels). This finding is confirmed by single-trial
beta weights (Fig. 8B, lower panels). As before, we restricted
the ANOVA to data from the left hemisphere. T2-P3 activity
was significantly modulated by SOA (F2,182 6.80, p .003,h2 .07, .83); post-hoc t-tests revealed that betas for SOA800 were smaller than for SOA 100 and for SOA 300 ( p < .001,
p .006; not corrected for multiple comparisons).
328 msec). In the right hemisphere, significant P3 related
indicate significant time points (deviation from zero). The
and T1LT2R trials, color-coded (red: positive, green: zero,
gle-trial level only in the left hemisphere. Ticks on the y-
s show average beta time courses in T1LT2R and T1RT2L
00 (blue). The upper left panel shows T2-P3 activity in the
the right hemisphere. In the left hemisphere, P3 can be
e right hemisphere, significant P3 related activity appears
cant time points (deviation from zero). The lower panels
T2R trials, color-coded (red: positive, green: zero, blue:
sphere. Small vertical bars on the x-axis indicate T2 onset
trials.
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Fig. 9 e Dual-task T2-P3 responses in the patient (left panel) and the control subjects (right panel, N [ 12). Shown are the
average T2-P3 beta time courses for SOA 100 (red), SOA 300 (green), and SOA 800 (blue) from T2 onset. Topographies
represent the RVF-P3 for the patient (latency: 328 msec) and the average of LVF-P3 and RVF-P3 for controls (latency:
380 msec). Missing values for SOA 800 (>.7 sec) are the result of EEG segmentation (L.2e1.5 sec from T1 onset). Theoffset at T2 onset (0 sec) is due to the fact that the T2-P3 beta weights were baseline corrected with respect to T1 onset.
The topographies are 2D representations of the different 3D electrode layouts (patient: 256 electrodes, Pz is the fifth
midline channel from the bottom; controls: 62 electrodes, Pz is the third midline channel from the bottom).
c o r t e x x x x ( 2 0 1 2 ) 1e1 814
Fig. 9 illustrates the significant difference in T2-
postponement between patient (left panel) and controls
(right panel) by plotting T2-P3 beta weights from T2 onset for
all SOAs. Contrary to previous findings in normal subjects
(Hesselmann et al., 2011; Sigman and Dehaene, 2008), the
patients T2-P3 showed only minimal latency postponement
at short SOAs during the PRP, an observation which was
corroborated by inspection of the raw averaged ERP ampli-
tudes in parietal electrodes (Fig. 3). For SOA 100, the T2-P3
latency was estimated at 340 msec relative to T2 onset, i.e.,
with a delay of 12 msec relative to T1-P3 latency in single-task
trials. At SOAs 300 and 800, the average T2-P3 peaks were
estimated at 332 msec (delay: 4 msec) and 344 msec (delay:
16 msec), respectively. In the healthy controls, the corre-
sponding T2-P3 delays were 280 msec (t11 5.90, p < .001),174 msec (t11 3.67, p < .005), and 8 msec (t11 < 1), and were ofcomparable magnitude as the behavioral PRP effects [(RT2
minus single-task RT): 299 msec at SOA 100, and 127 msec at
SOA 300]. Finally, a direct comparison revealed that beta
weights were significantly larger for T1-P3 than T2-P3 in the
patient (F1,276 10.87, p .001, h2 .04).
4. Discussion
In this study, we have probed the brain mechanisms under-
lying dual-task processing in a patient (AC) with posterior
callosal section using a lateralized number-comparison task.
Behaviorally, the patient appeared to show a PRP effect
superficially similar to that found in the healthy control group,
yet with important differences (SOA effect on RT1, RT slopes,
and crosstalk). At the neurophysiological level, we were able
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
to decompose single-task and dual-task ERPs into subcom-
ponents in the patient, and found larger N1 activity in the left
than in the right hemisphere and only low-amplitude to
nonexistent P3(b) activity in the right hemisphere in dual-task
trials. Furthermore, the P3 evoked by RVF targets was split,
showing a distinctly contralateral topography, and was not
postponed by the PRP effect, in strong contrast to the signifi-
cant T2-P3 postponement found in the healthy control group.
Each of these points is discussed in turn.
4.1. PRP and crosstalk effects
The behavioral data revealed that patient AC was able to
accurately perform the lateralized number-comparison task,
which is in agreement with previous reports of spared
smallerelarger magnitude comparison for Arabic numbers in
split-brain patients (Colvin et al., 2005; Seymour et al., 1994)
and in a patient with posterior callosal lesion (Cohen and
Dehaene, 1996). In contrast to the healthy control group, the
patients RTsweremodulated by visual field: responses to RVF
targets were longer than to LVF targets. An attentional bias
toward the left visual hemifield underlying the RT effect
seems highly unlikely, given the fact that the amplitudes of
the sensory N1 component were significantly larger for RVF
than for LVF targets. A left field/right hemisphere advantage
for number-comparison tasks has not been reported previ-
ously. However, this finding has to be interpreted with
caution, since it cannot be ruled out that the damage to the
patients left parietal lobe contributed to this effect.
In agreement with previous reports of robust PRP effects in
callosum-sectioned patients (Ivry et al., 1998; Ivry and
Hazeltine, 2000; Pashler et al., 1994), patient ACs dual-task
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 15
performance showeda lengthening of RT2and significantRT1-
RT2 correlations at short SOAs, which are considered as the
hallmarks of the PRP(Pashler, 1994; Pashler and Johnston,
1989). RT1 in T1LT2R trials was modulated by SOA to a larger
degree than in thehealthy sample, indicating that thepatients
right hemisphere was more sensitive than normal to SOA
effects; in addition, RT2 slopes observed for patient AC were
consistently steeper than in the control group. Analysis of
behavioral crosstalk effects in congruent and incongruent
trials revealedaneffect of target laterality:when thefirst target
was presented on the left (T1LT2R trials), crosstalkwas absent;
however, when the first target was presented on the right
(T1RT2L trials), we found significant crosstalk from T1 to T2 in
the RT2 data, which was of comparable magnitude as in the
controls, but only marginally significant backward crosstalk.
Thus, T2 processing was maximally insensitive to dual-task
interference when the second target was presented to the left
hemisphere (T1LT2R trials). A similar insensitivity of the left
hemisphere to inference has recently been reported in a split-
brain patient with callosal anarchic-hand syndrome (Verleger
et al., 2011). In a Simon task (Simon, 1969), the patients (GH)
left-hemispheric motor system remained insensitive against
interfering motor signals from the right-hemispheric motor
system, but not vice versa. Giesbrecht and Kingstone (2004)
investigated the behavioral performance of a split-brain
patient (JW) in the attentional blink (AB) paradigm, which is
closely related to the PRP (Wong, 2002; Jolicur, 1999; Arnell
and Duncan, 2002; Marti et al., 2012).Their study showed that
behavioral performance during the AB was more impaired
when T2 was presented to the right hemisphere.
It remains unclear to what degree strategical task moni-
toring might have affected the observed robust dual-task
interference in patient AC (Schumacher et al., 2001;
Hazeltine et al., 2008). However, the fact that response times
to T1 were slowed at short SOAs, considerably more than in
normal subjects, suggests that the dual-task slowing effect we
observed in the patient may indeed not have the same origin
as in normals. In healthy controls, the first task typically
unfolds almost without any influence of the presence of
a second stimulus at short or long SOA. The fact that patient
AC was, paradoxically, enormously slowed by a quasi-
simultaneous T2 stimulus in the other hemifield, is compat-
ible with the proposal that task prioritization may pose
a specific problem for callosal patientswho are trying to follow
the instruction of responding first to the stimulus that came
first (Hazeltine et al., 2008). According to this argument, the
superficial presence of a PRP effect in callosal patients, even
seemingly larger than in normals, may in fact hide a consid-
erable degree of parallel processing at the brain level
(Hazeltine et al., 2008). Indeed, our ERP data provided direct
evidence supporting this idea.
4.2. ERP decomposition of the PRP effect
In patient AC and in the control group, we found the sensory
N1 component to be fully stimulus-locked for T1 and T2, in
accordance with earlier ERP studies of the PRP effect (Sigman
and Dehaene, 2008; Brisson and Jolicur, 2007; Hesselmann
et al., 2011). Furthermore, we observed no attenuation of the
T2-N1 during the PRP. Together, these data provide further
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
evidence that perceptual processing occurs in parallel for T1
and T2 in our lateralized dual-task paradigm.
Analysis of P3 responses revealed a strictly contralateral
T1-P3 in thepatient, i.e., a subdivisionor splitting into LVF-P3
for LVF targets and RVF-P3 for RVF targets, unlike in normal
subjects where we found a single parietal P3 in the same task
(Hesselmann et al., 2011). Analysis of themoment-to-moment
correlations between ERP topographies in single-task T1L and
T1R trials corroborated this finding: while topographies
converged toward a shared pattern at the time of the P3 in
controls, the topographies remained highly dissimilar until
a much later time period in the patient (w600 msec). This
finding extends a previous report of ERP asymmetry in the
same patient, which showed that LVF and RVF targets evoke
highly different ERPs that remain largely contralateral for
a long period, between approximately 200 and 620 msec after
target onset (Cohenet al., 2000). Theobserved late convergence
toward a shared topography was probably mediated by intact
anterior commissural connections in patient AC, which have
been implicated in the transfer of stimulus-related semantic
activation in partial split-brain patients (Sidtis et al., 1981). Our
finding appears to be in good agreement with a conceptuali-
zation of P3 in which there are multiple cognitive processes
and neural generators underlying its generation (Johnson,
1993, 1986). This model has received confirmatory evidence
from intracranial recordings and EEG-fMRI studies showing
a highly distributed set of activated areas including hippo-
campus and temporal, parietal, and frontal association
cortices (Halgren et al., 1998; Mantini et al., 2009).
In the single-task condition, the LVF-P3 was absent inmost
trials but of comparable amplitude to the RVF-P3 when being
present. In dual-task trials, this difference was even more
expressed, and LVF-P3was virtually absent for T1 and T2. This
striking hemisphere difference cannot simply be accounted
for by the patients neurological status, since his right hemi-
sphere (supposedly the neural origin of the LVF-P3) was intact.
Speculatively, P3 attenuation in the right hemisphere might
be directly related to our finding of maximal behavioral dual-
task interference when the second target is presented to the
LVF. However, LVF-P3 was also virtually absent for T1 in dual-
task trials. Alternatively, our findings could suggest that the
P3, which reflects access to a central stage of distributed
processing associated with conscious report (Sergent et al.,
2005; Del Cul et al., 2007), might arise predominantly from
left-hemispheric networks when inter-hemispheric transfer
is fully or partly disrupted and the hemispheres are isolated.
At this stage, this possibility remains speculative, and
a conclusion on this point will have to await further research.
P3 asymmetries have previously been reported for split-brain
(Proverbio et al., 1994; Verleger et al., 2011; Kutas et al., 1990;
Satomi et al., 1995) and callosal agenesis patients (Bayard
et al., 2004), but the emerging picture is not yet conclusive.
In normal subjects, ERP studies have suggested that corpus
callosal size (as indexed by handedness) might be related to P3
amplitude and latency (Hoffman and Polich, 1999).
Similarly to the N1, both amplitude and latency of the T1-
P3 remained unaffected by SOA, in the patient and in
healthy controls. Latency of the T2-P3, however, showed
significant PRP-related postponement in controls
(Hesselmann et al., 2011), but remained remarkably time-
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 816
locked to T2 onset in patient AC. This observed absence of T2-
P3 postponement suggests greater parallel processing in
patient AC, afforded by the partial disconnection of the two
hemispheres. It is compatible with the existence of a tempo-
rary period during which two distinct states of global intra-
hemispheric communication (global workspaces) coexist
and permit partially parallel decision making, unlike normal
subjects in whom the presence of a global bi-hemispheric
state enforces serial processing of dual tasks (Hesselmann
et al., 2011; Dehaene and Naccache, 2001). Previous split-
brain studies have reported various cognitive processes that
can be performed in parallel by the left and right hemispheres
after commissurotomy, e.g., visual search for conjunction
targets in bilateral arrays (Luck et al., 1989, 1994).
The possibility of two separate global neuronal workspaces
in patient AC e strictly limited to the first few hundred milli-
seconds of stimulus processing e raises the question of the
quality of his subjective visual awareness during that time
period (Dehaene and Changeux, 2011). Conceivably, his
conscious awareness of the two targets might be delayed by
w600 msec, i.e., until a shared ERP topography on the scalp
indicated a successful interaction between higher-level post-
sensory decision processes in each hemisphere (Sidtis et al.,
1981). This hypothesis could be directly tested using
measures of quantified introspection (Marti et al., 2010; Corallo
et al., 2008). In normal subjects, related delays of temporal
selection, which diminish the conscious report of a target
stimulus, have already been shown for the AB (Vul et al., 2008).
More speculatively, there could exist a brief period with two
separate statesof visual awareness inpatientAC, limited to the
time between the completion of sensory processing and the
beginning of higher-level inter-hemispheric communication.
If, however, the P3 component is taken as an indicator of
conscious access (Del Cul et al., 2007; Gaillard et al., 2009;
Sergent et al., 2005), then its absence over the right hemi-
sphere for LVF targets, specifically during dual-task process-
ing, may indicate a single episode of conscious access
restricted to the RVF target (whether presented first or second),
with much delayed or reduced conscious access to the LVF
target. Again, these hypotheses could be evaluated through
quantified introspection (Marti et al., 2010; Corallo et al., 2008).
4.3. Concluding remarks
This study is the first to investigate PRP effects in a patient
with posterior callosal section using high-density EEG. We
have shown that ERP signatures of dual-task limitations are
affected when the two hemispheres are (partially) discon-
nected. The neurophysiological evidence suggests a consider-
able degree of parallel processing, even at the level of central
stages that are normally postponed and unfold strictly serially
in normal subjects. Follow-up studies could compare patient
ACs behavioral and ERP results to that from surgical split-
brain patients, and include single-trial measures of both P3
amplitude and latency to further explore conscious access in
isolated hemispheres (Dehaene and Changeux, 2011). Recent
advances in neuroimaging techniques also open up a fasci-
nating opportunity to relate PRP effects in lateralized tasks to
individual measures of inter-hemispheric communication in
normal subjects (Doron andGazzaniga, 2008; Genc et al., 2011).
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
Conflict of interest
None declared.
Acknowledgments
This experiment was supported by INSERM, CEA, and the
Human Frontiers Science Program. It constitutes part of
a general research program on functional neuroimaging of the
human brain which was sponsored by the Atomic Energy
Commission (DenisLeBihan).G.H.would like to thankMariano
Sigman for providing the original regression code,Marie-Anne
Henaff for information on the patient, Sebastien Marti for his
helpful comments on an earlier version of themanuscript, and
Guillaume Flandin for invaluable adhoc support.
Supplementary material
Supplementary data associated with this article can be found
in the online version, at doi:10.1016/j.cortex.2012.03.014.
r e f e r e n c e s
Arnell KM and Duncan J. Separate and shared sources of dual-task costs in stimulus identification and response selection.Cognitive Psychology, 44: 105e147, 2002.
Arnell KM, Helion AM, Hurdelbrink JA, and Pasieka B. Dissociatingsources of dual-task interference using humanelectrophysiology. Psychonomic Bulletin and Review, 11(1):77e83, 2004.
Banich MT. The missing link: The role of interhemisphericinteraction in attentional processing. Brain and Cognition, 36(2):128e157, 1998.
Bayard S, Gosselin N, Robert M, and Lassonde M. Inter- andintrahemispheric processing of visual event-related potentialsin the absence of the corpus callosum. Journal of CognitiveNeuroscience, 16(3): 401e414, 2004.
Bell AJ and Sejnowski TJ. An information maximisation approachto blind separation and blind deconvolution. NeuralComputation, 7: 1129e1159, 1995.
Blair RC and Karniski W. An alternative method for significancetesting of waveform difference potentials. Psychophysiology,30(5): 518e524, 1993.
Brisson B and Jolicur P. Electrophysiological evidence of centralinterference in the control of visuospatial attention.Psychonomic Bulletin and Review, 14(1): 126e132, 2007.
Cohen J. Statistical Power Analysis for the Behavior Sciences. 2nd ed.Hillsdale, NJ: Erlbaum, 1988.
Cohen L and Dehaene S. Cerebral networks for numberprocessing: Evidence from a case of posterior callosal lesion.Neurocase, 2: 155e174, 1996.
Cohen L, Dehaene S, Naccache L, Lehericy S, Dehaene-Lambertz G, Henaff MA, et al. The visual word form area:Spatial and temporal characterization of an initial stage ofreading in normal subjects and posterior split-brain patients.Brain, 123(2): 291e307, 2000.
Colvin MK, Funnell MG, and Gazzaniga MS. Numerical processingin the two hemispheres: Studies of a split-brain patient. Brainand Cognition, 57(1): 43e52, 2005.
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 8 17
Corallo G, Sackur J, Dehaene S, and Sigman M. Limits onintrospection: Distorted subjective time during the dual-taskbottleneck. Psychological Science, 19(11): 1110e1117, 2008.
Crawford JR and Howell DC. Comparing an individuals test scoreagainst norms derived from small samples. The ClinicalNeuropsychologist, 12(4): 482e486, 1998.
De Jong R. Multiple bottlenecks in overlapping task performance.Journal of Experimental Psychology: Human Perception andPerformance, 19(5): 965e980, 1993.
Dehaene S and Changeux JP. Experimental and theoreticalapproaches toconsciousprocessing.Neuron, 70(2): 200e227, 2011.
Dehaene S and Naccache L. Towards a cognitive neuroscience ofconsciousness: Basic evidence and a workspace framework.Cognition, 79(1e2): 1e37, 2001.
Dehaene S, Naccache L, Le ClecH G, Koechlin E, Mueller M,Dehaene-Lambertz G, et al. Imaging unconscious semanticpriming. Nature, 395(6702): 597e600, 1998.
Del Cul A, Baillet S, and Dehaene S. Brain dynamics underlyingthe nonlinear threshold for access to consciousness. PLoSBiology, 5(10): 1e16, 2007.
DellAcqua R, Jolicur P, Vespignani F, and Toffanin P. Centralprocessing overlap modulates P3 latency. Experimental BrainResearch, 165(1): 54e68, 2005.
Delorme A and Makeig S. EEGlab: An open source toolbox foranalysis of single-trial EEG dynamics. Journal of NeuroscienceMethods, 134(1): 9e21, 2004.
Donchin E. Surprise!.Surprise? Psychophysiology, 18(5): 493e513, 1981.Donchin E and Coles MG. Is the P300 component a manifestation
of context updating? Behavioral and Brain Sciences, 11(3):357e374, 1988.
Doron KW and Gazzaniga MS. Neuroimaging techniques offernew perspectives on callosal transfer and interhemisphericcommunication. Cortex, 44(8): 1023e1029, 2008.
Frak V, Paulignan Y, Jeannerod M, Michel F, and Cohen H.Prehension movements in a patient (AC) with posteriorparietal cortex damage and posterior callosal section. Brainand Cognition, 60(1): 43e48, 2006.
Franz EA, Eliassen JC, Ivry RB, and Gazzaniga MS. Dissociation ofspatial and temporal coupling in the bimanual movements ofcallosotomy patients. Psychological Science, 7(5): 306e310, 1996.
Gaillard R, Dehaene S, Adam C, Clemenceau S, Hasboun D,Baulac M, et al. Converging intracranial markers of consciousaccess. PLoS Biology, 7: e61, 2009.
Gazzaniga MS. Principles of human brain organization derivedfrom split-brain studies. Neuron, 14(2): 217e228, 1995.
Gazzaniga MS. Forty-five years of split-brain research and stillgoing strong. Nature Reviews Neuroscience, 6(8): 653e659, 2005.
Genc E, Bergmann J, Singer W, and Kohler A. Interhemisphericconnections shape subjective experience of bistable motion.Current Biology, 21(17): 1494e1499, 2011.
Giesbrecht B and Kingstone A. Right hemisphere involvement inthe attentional blink: Evidence from a split-brain patient. Brainand Cognition, 55(2): 303e306, 2004.
Greenhouse SW and Geisser S. On methods in the analysis ofprofile data. Psychometrika, 24(2): 95e112, 1959.
Groppe DM, Urbach TP, and Kutas M. Mass univariate analysis ofevent-related brain potentials/fields I: A critical tutorialreview. Psychophysiology, 48(12): 1711e1725, 2011.
Halgren E, Marinkovic K, and Chauvel P. Generators of the latecognitive potentials in auditory and visual oddball tasks.Electroencephalography andClinicalNeurophysiology, 106(2): 156e164,1998.
Hazeltine E, Weinstein A, and Ivry RB. Parallel response selectionafter callosotomy. Journal of Cognitive Neuroscience, 20(3):526e540, 2008.
Hesselmann G, Flandin G, and Dehaene S. Probing the corticalnetwork underlying the psychological refractory period: Acombined EEG-fMRI study. NeuroImage, 56(3): 1608e1621, 2011.
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
Hofer S and Frahm J. Topography of the human corpus callosumrevisited e Comprehensive fiber tractography using diffusiontensor magnetic resonance imaging. NeuroImage, 32(3):989e994, 2006.
Hoffman LD and Polich J. P300, handedness, and corpus callosalsize: Gender, modality, and task. International Journal ofPsychophysiology, 31(2): 163e174, 1999.
Holtzman JD and Gazzaniga MS. Dual task interactions dueexclusively to limits in processing resources. Science, 218(4579):1325e1327, 1982.
Intriligator J, Henaff MA, and Michel F. Able to name, unable tocompare: The visual abilities of a posterior split-brain patient.NeuroReport, 11(12): 2639e2642, 2000.
Isreal JB, Chesney GL, Wickens CD, and Donchin E. P300 andtracking difficulty: Evidence for multiple resources in dual-task performance. Psychophysiology, 17(3): 259e273, 1980.
Ivry R and Hazeltine E. Task switching in a callosotomy patientand normal participants: Evidence for response-relatedsources of interference. In Monsell R and Driver J (Eds)Attention and Performance XVIII. Cambridge, MA: MIT Press,2000: 401e423.
Ivry RB, Franz EA, Kingstone A, and Johnston JC. Thepsychological refractory period effect following callosotomy:Uncoupling of lateralized response codes. Journal ofExperimental Psychology: Human Perception and Performance,24(2): 463e480, 1998.
Johannes S, Munte TF, Heinze HJ, and Mangun GR. Luminanceand spatial attention effects on early visual processing.Cognitive Brain Research, 2(3): 189e205, 1995.
Johnson Jr R. A triarchic model of P300 amplitude.Psychophysiology, 23(4): 367e384, 1986.
Johnson Jr R. On the neural generators of the P300 component ofthe event-related potential. Psychophysiology, 30(1): 90e97,1993.
Jolicur P. Concurrent response-selection demands modulate theattentional blink. Journal of Experimental Psychology: HumanPerception and Performance, 25(4): 1097e1113, 1999.
Kok A. On the utility of P3 amplitude as a measure of processingcapacity. Psychophysiology, 38(3): 557e577, 2001.
Kutas M, Hillyard SA, Volpe BT, and Gazzaniga MS. Late positiveevent-related potentials after commissural sections inhumans. Journal of Cognitive Neuroscience, 2(3): 258e271, 1990.
Lehmann D and Skrandies W. Reference-free identification ofcomponents of checkerboard-evoked multichannel potentialfields. Electroencephalography and Clinical Neurophysiology, 48(6):609e621, 1980.
Logan GD and Delheimer JA. Parallel memory retrieval in dual-task situations: II. Episodic memory. Journal of ExperimentalPsychology: Learning, Memory and Cognition, 27(3): 668e685, 2001.
Logan GD and Schulkind MD. Parallel memory retrieval in dual-task situations: I. Semantic memory. Journal of ExperimentalPsychology: Human Perception and Performance, 26(3): 1072e1090,2000.
Luck SJ. Sources of dual-task interference: Evidence from humanelectrophysiology. Psychological Science, 9(3): 223e227, 1998.
Luck SJ, Hillyard SA, Mangun GR, and Gazzaniga MS. Independenthemispheric attentional systems mediate visual search insplit-brain patients. Nature, 342(6249): 543e545, 1989.
Luck SJ, Hillyard SA, Mangun GR, and Gazzaniga MS. Independentattentional scanning in the separated hemispheres of split-brain patients. Journal of Cognitive Neuroscience, 6(1): 84e91,1994.
Mantini D, Corbetta M, Perrucci MG, Romani GL, and Del Gratta C.Large-scale brain networks account for sustained andtransient activity during target detection. NeuroImage, 44(1):265e274, 2009.
Marois R and Ivanoff J. Capacity limits of information processingin the brain. Trends in Cognitive Sciences, 9(6): 296e305, 2005.
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
c o r t e x x x x ( 2 0 1 2 ) 1e1 818
Marti S, Sackur J, Sigman M, and Dehaene S. Mappingintrospections blind spot: Reconstruction of dual-taskphenomenology using quantified introspection. Cognition,115(2): 303e313, 2010.
Marti S, Sigman M, and Dehaene S. A shared cortical bottleneckunderlying attentional blink and psychological refractoryperiod. NeuroImage, 59(3): 2883e2898, 2012.
Michel F, Henaff MA, and Intriligator J. Two different readers inthe same brain after a posterior callosal lesion. NeuroReport,7(3): 786e788, 1996.
Miller J. Backward crosstalk effects in psychological refractoryperiod paradigms: Effects of second-task response types onfirst-task response latencies. Psychological Research, 70(6):484e493, 2006.
Mognon A, Jovicich J, Bruzzone L, and Buiatti M. ADJUST: Anautomatic EEG artifact detector based on the joint use of spatialand temporal features. Psychophysiology, 48(2): 229e240, 2011.
Naccache L and Dehaene S. Unconscious semantic primingextends to novel unseen stimuli. Cognition, 80(3): 215e229, 2001.
Navon D and Miller J. Queuing or sharing? A critical evaluation ofthe single-bottleneck notion. Cognitive Psychology, 44(3):193e251, 2002.
Oostenveld R, Fries P, Maris E, Schoffelen JM. FieldTrip: Opensource software for advanced analysis of MEG, EEG, andinvasive electrophysiological data. Computational Intelligenceand Neuroscience, 1e9, 2011. doi:10.1155/2011/156869.
Pashler H. Dual-task interference in simple tasks: Data andtheory. Psychological Bulletin, 116(2): 220e244, 1994.
Pashler H. The Psychology of Attention. Cambridge: MA MIT Press,1998.
Pashler H and Johnston JC. Interference between temporallyoverlapping tasks: Chronometric evidence for centralpostponement with or without response grouping. QuarterlyJournal of Experimental Psychology, 41A(1): 19e45, 1989.
Pashler H, Luck SJ, Hillyard SA, Mangun GR, OBrien S, andGazzaniga MS. Sequential operation of disconnected cerebralhemispheres in split-brain patients. NeuroReport, 5(17):2381e2384, 1994.
Picton TW. The P300 wave of the human event-related potential.Journal of Clinical Neurophysiology, 9(4): 456e479, 1992.
Proverbio AM, Zani A, Gazzaniga MS, and Mangun GR. ERP and RTsigns of a rightward bias for spatial orienting in a split-brainpatient. NeuroReport, 5(18): 2457e2461, 1994.
Sackur J, Naccache L, Pradat-Diehl P, Azouvi P, Mazevet D, Katz R,et al. Semantic processing of neglected numbers. Cortex, 44(6):673e682, 2008.
Satomi K, Horai T, Kinoshita Y, and Wakazono A. Hemisphericasymmetry of event-related potentials in a patient withcallosal disconnection syndrome: A comparison of auditory,
Please cite this article in press as: Hesselmann G, et al., Splittingwith posterior callosal section, Cortex (2012), doi:10.1016/j.cortex
visual and somatosensory modalities. Electroencephalographyand Clinical Neurophysiology, 94(6): 440e449, 1995.
Scalf PE, Banich MT, Kramer AF, Narechania K, and Simon CD.Double take: Parallel processing by the cerebral hemispheresreduces attentional blink. Journal of Experimental Psychology:Human Perception and Performance, 33(2): 298e329, 2007.
Schumacher EH, Seymour TL, Glass JM, Fencsik DE, Lauber EJ,Kieras DE, et al. Virtually perfect time sharing in dual-taskperformance: Uncorking the central cognitive bottleneck.Psychological Science, 12(2): 101e108, 2001.
Sergent C, Baillet S, and Dehaene S. Timing of the brain eventsunderlying access to consciousness during the attentionalblink. Nature Neuroscience, 8(10): 1391e1400, 2005.
Seymour SE, Reuter-Lorenz PA, and Gazzaniga MS. Thedisconnection syndrome. Basic findings reaffirmed. Brain,117(1): 105e115, 1994.
Sidtis JJ, Volpe BT, Holtzman JD, Wilson DH, and Gazzaniga MS.Cognitive interaction after staged callosal section: Evidence fortransfer of semantic activation. Science, 212(4492): 344e346, 1981.
Sigman M and Dehaene S. Parsing a cognitive task: Acharacterization of the minds bottleneck. PLoS Biology, 3(2):e37, 2005.
Sigman M and Dehaene S. Brain mechanisms of serial andparallel processing during dual-task performance. Journal ofNeuroscience, 28(30): 7585e7598, 2008.
Simon JR. Reactions toward the source of stimulation. Journal ofExperimental Psychology, 81(1): 174e176, 1969.
Telford CW. The refractory phase of voluntary and associativeresponses. Journal of Experimental Psychology, 14(1): 1e36, 1931.
Todman JB and Dugard P. Single-case and Small-n ExperimentsDesigns. New York: Lawrence Erlbaum Associates, 2001.
Tombu M and Jolicur P. A central capacity sharing model ofdual-task performance. Journal of Experimental Psychology:Human Perception and Performance, 29(1): 3e18, 2003.
Verleger R, Binkofski F, Friedrich M, Sedlmeier P, and Kompf D.Anarchic-hand syndrome: ERP reflections of lost control overthe right hemisphere. Brain and Cognition, 77(1): 138e150, 2011.
Verleger R, Jaskowski P, and Wascher E. Evidence for anintegrative role of P3b in linking reaction to perception. Journalof Psychophysiology, 19(3): 165e181, 2005.
Vul E, Nieuwenstein M, and Kanwisher N. Temporal selection issuppressed, delayed, and diffused during the attentionalblink. Psychological Science, 19(1): 55e61, 2008.
Welford AT. The psychological refractory period and the timingof high-speed performance e a review and a theory. BritishJournal of Psychology, 43(1): 2e19, 1952.
Wong KFE. The relationship between attentional blink andpsychological refractory period. Journal of Experimental Psychology:Human Perception and Performance, 28(1): 54e71, 2002.
of the P3 component during dual-task processing in a patient.2012.03.014
http://dx.doi.org/10.1155/2011/156869http://dx.doi.org/10.1016/j.cortex.2012.03.014http://dx.doi.org/10.1016/j.cortex.2012.03.014
Splitting of the P3 component during dual-task processing in a patient with posterior callosal section1. Introduction1.1. Dual-task performance in split-brain patients1.2. The PRP1.3. Event-related potential (ERP) studies of the PRP
2. Materials and methods2.1. Participants2.2. Design and procedure2.3. Behavioral data analysis2.4. EEG acquisition and preprocessing2.5. EEG analysis
3. Results3.1. Behavioral results: PRP effect3.2. Behavioral results: crosstalk effects3.3. EEG results: ERPs in single-task and dual-task trials3.4. EEG results: single-task (N1 and P3)3.5. EEG results: dual-task (N1)3.6. EEG results: dual-task (P3)
4. Discussion4.1. PRP and crosstalk effects4.2. ERP decomposition of the PRP effect4.3. Concluding remarks
Conflict of interestAcknowledgmentsAppendix. Supplementary materialReferences